Key Technologies Research On The High Frequency Surface Wave Radar Echo Simulator Based On The GPU

Performance test and algorithm verification are indispensable in the process of radar system evaluation.In order to obtain accurate evaluation results,radar echo simulator is used to test the availability and stability of signal processing algorithms.Due to the obvious shortcoming of the echo generation rate and simple target motion model in the traditional CPU-based echo simulator which is used in the HFSWR,in this paper a high performance echo simulate method is proposed.As the observation of HFSWR is very huge,the echo is composed of multi-targets,radio,interference and clutter which need high sampling rate.In the traditional CPU-based platform,the consumption of time for generating multi-target radar echo is very high which cannot afford real-time requirement.So the performance evaluation and testing efficiency of HFSWR is low.With the help of the tremendous parallel computational horsepower in GPU,it can be used for HFSWR real-time multi-target echo simulation.This thesis focuses on the parallel realization technique of multi-target echo simulation of HFSWR.The major contributions of this dissertation are as follows:First of all,we establish the complex motion model of target movi ng.The target motion information of an uniformly accelerated motion in any direction can be achieved by this model which can be used for the following echo simulation.Secondly,the basic principles of HFSWR echo simulation is explored in detail.By exploring the parallelism that exists in the echo simulation,it is possible of GPU to be used for HFSWR echo simulation.Thirdly,the realization scheme of echo simulator is designed and the echo simulator is realized by C# and CUDA multi-language programming technology.Taking the advantage of interfacial design of C#,we accomplish the interfacial design of echo simulator.In this interfacial,we can set the system parameters and target parameters.We also accomplish the parallelism realization of baseband si gnal and target motion information by CUDA.In the end,by the fusion of target motion information and radar digital echo signal,we accomplish the generation of multi-target HFSWR echo data in real-time.By comparing with the traditional CPU-based simulator,the speed-up ratio of the time to generate equivalent single target radar echo is 200 which can generate radar echo data of 200 targets in 4.28 s which the pulse accumulation time is 8.448 s.So the real-time requirement is satisfied which improves performance evaluation and testing efficiency of HFSWR.